Abstract
BACKGROUND: Alzheimer's disease (AD) is a progressive neurodegenerative disorder, with mild cognitive impairment (MCI) often serving as its precursor stage. Early intervention at the MCI stage can significantly delay AD onset. METHODS: This study employed untargeted urine metabolomics, with data obtained from the MetaboLights database (MTBLS8662), combined with orthogonal partial least squares-discriminant analysis (OPLS-DA) to examine metabolic differences across different stages of AD progression. A decision tree approach was used to identify key metabolites within significantly enriched pathways. These key metabolites were then utilized to construct and validate an AD progression prediction model. RESULTS: The OPLS-DA model effectively distinguished the metabolic characteristics at different stages. Pathway enrichment analysis revealed that Drug metabolism was significantly enriched across all stages, while Retinol metabolism was particularly prominent during the transition stages. Key metabolites such as Theophylline, Vanillylmandelic Acid (VMA), and Adenosine showed significant differencesdifferencesin the early stages of the disease, whereas 1,7-Dimethyluric Acid, Cystathionine, and Indole exhibited strong predictive value during the MCI to AD transition. These metabolites play a crucial role in monitoring AD progression. Predictive models based on these metabolites demonstrated excellent classification and prediction capabilities. CONCLUSION: This study systematically analyzed the dynamic metabolic differences during the progression of AD and identified key metabolites and pathways as potential biomarkers for early prediction and intervention. Utilizing urinary metabolomics, the findings provide a theoretical basis for monitoring AD progression and contribute to improving prevention and intervention strategies, thereby potentially delaying disease progression.